Due Diligence Capital Notes

CleanSpark just reported a $757 million net loss for the first half of fiscal 2026. In the same six-month window, it spent $460 million buying back its own stock. Read that sentence again — a company bleeding three-quarters of a billion dollars on paper is simultaneously executing one of the most aggressive share repurchase programs in the Bitcoin mining sector. That tension is exactly what makes CleanSpark worth understanding right now.

The loss was not the result of a failed business. Mining revenue came in at $317.6 million for the half-year. CleanSpark added nearly 1,500 coins to its treasury, ending March 31, 2026 with 11,920 BTC — up from 10,428 at the start of the fiscal year. The loss came from an accounting rule that forces the company to write down its Bitcoin holdings every quarter based on the prevailing market price. When Bitcoin fell from $114,068 per coin to $68,222 per coin during the period, the numbers on CleanSpark's income statement got ugly fast, even though not a single coin was actually sold. Understanding why that distinction matters — and what it tells us about CleanSpark's underlying capital allocation bet — is the core of this post.

ASC 820 and the Mark-to-Market Paradox

ASC 820 is the U.S. accounting standard (issued by the Financial Accounting Standards Board) that requires companies to report certain assets — including Bitcoin, after a 2023 update — at fair value, meaning the current market price, every single reporting period. The gain or loss from that price movement flows directly through the income statement whether or not the company has sold anything.

This creates a paradox for Bitcoin miners that accumulate rather than sell. If you hold 11,920 BTC and the price drops $45,846 per coin over six months, your income statement absorbs a $470.9 million hit regardless of your operational performance. That is precisely what happened to CleanSpark in the first half of fiscal 2026. The fair value of its Bitcoin treasury dropped from an implied high near $1.36 billion to $813.2 million — a paper loss that accounts for the majority of the $757.1 million net loss reported for the period.

The accounting rule is theoretically symmetrical: when Bitcoin rises, the same standard produces paper gains that inflate reported earnings. But symmetry in accounting does not make the volatility comfortable for investors who focus on headline net income without reading the footnotes. This is why understanding why CleanSpark is losing money matters more than that it is losing money.

The Moving Parts of CleanSpark's Capital Machine

CleanSpark's strategy has several distinct levers running simultaneously, and each one carries its own logic and risk profile.

Self-mined Bitcoin accumulation as a treasury strategy. Rather than selling Bitcoin immediately after mining it — the model most public miners used in prior cycles — CleanSpark is retaining coins on its balance sheet. As of March 31, 2026, it held 11,920 BTC with a total fair value of $813.2 million. The average cost basis per coin sits at $97,309, meaning the treasury is currently underwater relative to both the cost to acquire and the peak price of $114,068 seen earlier in the reporting period. Crucially, 96.6% of those coins are stored in cold wallets — meaning they are held offline, away from exchange custody, reducing counterparty risk at the cost of liquidity. The company has also disclosed that it uses bitcoin-linked derivative contracts — financial instruments whose value is tied to Bitcoin's price — to hedge against short-term price drops and generate liquidity without outright sales. As the 10-Q filed on May 11, 2026 explains: "These contracts serve as a strategic alternative to selling bitcoin directly and are intended to monetize the Company's bitcoin holdings while managing exposure to adverse price movements."
Debt-funded infrastructure buildout. CleanSpark raised $1.13 billion in new debt during the first half of fiscal 2026 alone, pushing long-term debt — net of issuance costs — from $644.6 million to $1.79 billion. That is almost a tripling of the debt load in six months. The proceeds funded land acquisitions in Texas (Brazoria County for $28.7 million, Austin County for $65.7 million), outstanding miner deposits totaling $137.4 million, and the expansion of active data centers across Georgia, Tennessee, Mississippi, and Wyoming, with South Dakota and Texas still in development. The cash balance surged from $43 million to $260.3 million, almost entirely because of these debt issuances. This is a company borrowing aggressively to build physical infrastructure at scale — a bet that higher hashrate capacity, once online, will produce more Bitcoin than the interest on those convertible notes costs.
Aggressive share buybacks. This is the most unusual element of CleanSpark's capital allocation, and the one that most directly signals management's view of the stock's intrinsic value. In the first half of fiscal 2026, the company repurchased $460 million of its own common shares. The total treasury stock — shares bought back and held off the market — now stands at $608.2 million across 42.4 million shares. With 256,608,606 shares outstanding as of May 7, 2026, those repurchases represent a meaningful reduction in the share count. Buybacks work by concentrating ownership: if the pie stays the same size but gets cut into fewer slices, each remaining slice is worth more. The logic only holds if management is buying below intrinsic value. In this case, management is effectively arguing that CLSK shares are cheap even as the balance sheet absorbs a nine-figure paper loss — a high-conviction signal that is either bold or reckless, depending on how Bitcoin prices move from here.
Depreciation as a hidden cash flow cushion. CleanSpark recorded $222.2 million in total depreciation and amortization during the half-year, with $199.5 million attributable to miner depreciation alone. Depreciation is a non-cash accounting charge — it reduces reported income but does not require any cash to leave the business. When you add back $222.2 million in D&A to the $757.1 million net loss, the operational cash picture looks substantially less dire than the headline number. This does not erase the debt problem, but it explains how a company with massive reported losses can continue funding expansion and buybacks simultaneously.
The HPC and AI pivot signal. CleanSpark's risk factor disclosures in the same 10-Q explicitly flag "our ability to diversify and expand into the market for high-performance computing (HPC) and artificial intelligence (AI) solutions and data centers" as a key variable. HPC refers to computing infrastructure used for tasks requiring massive parallel processing — including the model training and inference workloads that hyperscaler companies like Microsoft, Amazon, and Google are frantically building capacity for. The Texas land acquisitions in Brazoria and Austin counties are geographically consistent with a pivot toward serving those customers. This is the same playbook that companies like Core Scientific have begun executing — repurposing or co-locating mining infrastructure alongside AI data center services to diversify revenue beyond Bitcoin's price cycle.

The Numbers That Deserve a Closer Look

Mining revenue declined year-over-year despite a larger fleet: $317.6 million in H1 FY2026 versus $344.0 million in the prior-year period. That gap reflects the April 2024 halving — the programmatic event that cut the Bitcoin reward per block from 3.125 BTC to fewer coins — combined with a higher global network difficulty. More miners in the field competing for the same block rewards means each individual machine earns less Bitcoin per hour. CleanSpark's fleet is getting bigger, but the unit economics of Bitcoin mining are compressing across the entire industry.

The accumulated deficit stands at $912.9 million. Accumulated deficit is the running total of all net losses a company has recorded since inception, minus any net profits. It is not a cash figure — it does not mean the company owes that money to anyone — but it does reflect that CleanSpark has never had a profitable fiscal year on a GAAP basis. For a company now carrying $1.79 billion in long-term debt with convertible notes maturing in 2030 and 2032, the path to covering those obligations runs almost entirely through Bitcoin price appreciation and the eventual monetization of its infrastructure assets.

There was also a $4.0 million write-off of a transformer deposit and $1.4 million in idle equipment charges during the half — small numbers in isolation, but a reminder that rapid physical infrastructure build-outs generate friction costs that never appear in the headline metrics.

What Could Break This Thesis

Bitcoin price stagnation or decline. This is the single biggest risk, and the filing period demonstrates it viscerally. A decline from $114,068 to $68,222 per coin produced a $470.9 million fair value loss in six months under ASC 820. With a cost basis of $97,309 per coin, CleanSpark's entire treasury is currently underwater. If Bitcoin trades sideways or lower for an extended period, the company will continue reporting massive net losses while its debt obligations compound — a dangerous combination.

Debt service capacity under operational stress. Long-term debt nearly tripled in six months. The convertible notes due in 2030 and 2032 must eventually be repaid or refinanced. If Bitcoin enters a multi-year bear market and mining economics deteriorate further, the company's ability to service those obligations — or refinance them at reasonable rates — becomes genuinely uncertain. Cash of $260.3 million against $1.79 billion in debt is not a comfortable cushion if revenue weakens materially.

Tariff risk on miner procurement. The 10-Q explicitly warns of "uncertainty as to whether the Company will face materially increased tariff liability in respect of miners purchased since 2024 and in the future." Bitcoin miners are manufactured primarily in Asia, and changes in U.S. trade policy could significantly increase the capital expenditure required to maintain and expand the fleet. With $137.4 million in outstanding miner deposits already committed, any tariff escalation hits margins before the machines even arrive.

HPC and AI execution risk. CleanSpark openly acknowledges in its forward-looking statements that it has "limited experience" in HPC and AI markets and must "secure hyperscaler customers" against entrenched competitors who have been building these relationships for years. Pivoting from Bitcoin mining — where the customer is the Bitcoin network itself — to selling data center capacity to enterprise hyperscalers requires a fundamentally different go-to-market capability. Announcing the intention is not the same as delivering the revenue.

Conclusion

CleanSpark is making a high-conviction, multi-year capital allocation bet: borrow aggressively, build physical infrastructure, accumulate Bitcoin, and reduce the share count — all simultaneously. The strategy presupposes that Bitcoin prices will recover and appreciate from the current $68,222 per coin level, that the Texas data center buildout will attract either mining revenue or HPC contracts at attractive margins, and that the $1.79 billion debt load can be managed or refinanced before the 2030 maturity arrives.

What I find genuinely interesting is the combination of buybacks and accumulation. Management is not hedging its conviction — it is doubling down on both the asset (Bitcoin) and the equity (CLSK shares) at the same time. That is either a disciplined expression of long-term value thinking or a dangerous concentration of risk. Given that the cost basis on the Bitcoin treasury sits at $97,309 per coin against a current market price of roughly $68,222, the thesis is not yet working. But the infrastructure is being built for a world where it does — and CleanSpark's willingness to absorb that paper pain without selling a single coin is the clearest possible signal about where management expects Bitcoin to be when those 2030 notes come due.